JP2010209829A - Air-fuel ratio control device for internal combustion engine - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To suppress a reduction in learning accuracy caused by learning of a feedback learning value due to the update of a first learning value in driving from the fully open state of a throttle valve to a closing side, and to suppress a reduction in the learning frequency of the feedback learning value. <P>SOLUTION: A sub feedback learning value SG is calculated by using a formula, "SG=B+(A-B)×K" where A represents the first learning value, B represents a second learning value and K represents an interpolation correction value. The interpolation correction value K is reduced so that the sub feedback learning value SG is reduced according as the maximum lift and operating angle of an intake valve 9 are increased and the intake pressure of an engine 1 is reduced. When the interpolation correction value K is a determining value J or more, the corresponding first learning value A when the maximum lift and operating angle are increased is updated, and when the interpolation correction value K is below the determining value J, the corresponding second learning value B when the maximum lift and operating angle are reduced is updated. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an air-fuel ratio control apparatus for an internal combustion engine.

  In an internal combustion engine such as a multi-cylinder engine for automobiles, the maximum lift amount of the intake valve by the variable valve lift mechanism is set while opening the throttle valve provided in the intake passage as much as possible to improve fuel efficiency by reducing pumping loss. Further, there is known a system in which the intake air amount of the engine is adjusted through variable operation angle. However, even in such an internal combustion engine, when the required amount of intake air amount of the internal combustion engine is small, such as during low-load operation, the intake air amount can be reduced only by changing the maximum lift amount and operating angle of the intake valve by the variable valve lift mechanism. Since the amount cannot be reduced to the required amount, the throttle valve is driven from the fully open position to the closed side to reduce the intake air amount to the required amount. In addition, even when the maximum lift amount and operating angle of the intake valve cannot be reduced due to the sticking abnormality of the variable valve lift mechanism, the intake air amount of the internal combustion engine cannot be reduced to the required amount. The intake air amount is reduced to the required amount by driving to.

  When the intake air amount of the internal combustion engine is adjusted as described above, the fuel injection from the fuel injection valve is performed in consideration of the air-fuel ratio of the engine based on the intake air amount and the like, and consists of air and fuel. The internal combustion engine is driven by the combustion of the air-fuel mixture. Regarding exhaust after combustion of fuel in an internal combustion engine, oxidation of CO, HC and reduction of NOx in the exhaust is performed by a three-way catalyst of a catalytic converter provided in an exhaust passage, and these are harmless CO2, H2O, Purification is achieved by using N2. Exhaust gas purification by such a three-way catalyst, that is, oxidation of CO and HC and reduction of NOx is most effectively performed at the oxygen concentration of the catalyst atmosphere when the air-fuel mixture is burned at the stoichiometric air-fuel ratio. For this reason, in the internal combustion engine, air-fuel ratio feedback control is performed in which the actual air-fuel ratio of the engine is the stoichiometric air-fuel ratio.

  Air-fuel ratio feedback control in which the actual air-fuel ratio of the internal combustion engine is the stoichiometric air-fuel ratio uses a sensor that is provided in the exhaust passage and outputs a signal corresponding to the oxygen concentration in the exhaust gas. This is realized by correcting the injection amount.

  Specifically, the fuel injection of the internal combustion engine is performed using a feedback correction value that increases or decreases based on a deviation from a target value that is a value obtained when the actual air-fuel ratio of the engine at the same output as the theoretical air-fuel ratio at the same output. The amount is corrected. That is, when the output of the sensor is a value on the rich side with respect to the target value, the feedback correction value is reduced and the fuel injection amount is corrected to decrease based on the correction value, and the output of the sensor is lean relative to the target value. If the value is a value on the side, the feedback correction value is increased, and the fuel injection amount is corrected to increase based on the correction value. Through the increase / decrease correction of the fuel injection amount based on the feedback correction value, an instantaneous deviation of the actual air-fuel ratio from the stoichiometric air-fuel ratio in the internal combustion engine is suppressed, and the actual air-fuel ratio is controlled to the stoichiometric air-fuel ratio.

  In addition, manufacturing variations and changes with time occur in the intake system and fuel supply system of the internal combustion engine, and as a result, a steady deviation of the actual air-fuel ratio of the engine from the stoichiometric air-fuel ratio occurs. The feedback learning value is learned so as to be a value corresponding to the steady deviation based on the feedback correction value, with the intention of compensating for the steady deviation of the actual air-fuel ratio from the theoretical air-fuel ratio. The fuel injection amount of the internal combustion engine is also corrected using the learned value. Then, through the correction of the fuel injection amount based on the feedback learning value, the steady deviation of the actual air fuel ratio from the theoretical air fuel ratio is compensated.

  As described above, the feedback correction value and the feedback learning value are obtained, and the fuel injection amount is corrected based on the feedback correction value, thereby realizing air-fuel ratio feedback control in which the actual air-fuel ratio of the internal combustion engine is the stoichiometric air-fuel ratio.

  By the way, in an internal combustion engine that adjusts the intake air amount by varying the maximum lift amount and the operating angle of the intake valve by the variable valve lift mechanism, as a cause of steady deviation of the actual air fuel ratio from the theoretical air fuel ratio, It is also possible that the passage area of the connection portion of the intake passage with the combustion chamber deviates from an appropriate value due to assembly errors and deposits deposited on the valve. The steady deviation of the actual air-fuel ratio from the theoretical air-fuel ratio due to such a cause occurs when the maximum lift amount and operating angle of the intake valve are small.

  This is because when the maximum lift amount and operating angle of the intake valve are small, the intake air amount of the internal combustion engine decreases, and therefore, the deviation amount with respect to the appropriate value of the intake air amount of the engine due to the above cause occupies the entire intake air amount. This is because the ratio increases and this appears as a steady deviation of the actual air-fuel ratio from the stoichiometric air-fuel ratio. On the other hand, since the intake air amount of the internal combustion engine increases when the maximum lift amount and the operating angle of the intake valve are large, the ratio of the deviation amount to the appropriate value of the intake air amount of the engine due to the above cause to the entire intake air amount Becomes smaller and does not appear as a steady deviation of the actual air-fuel ratio with respect to the theoretical air-fuel ratio.

  Compensation can also be achieved for the steady deviation of the actual air-fuel ratio of the internal combustion engine from the above cause with respect to the stoichiometric air-fuel ratio by learning the feedback learning value and correcting the fuel injection amount based on the learning value. is there.

  However, in an internal combustion engine that adjusts the intake air amount by varying the maximum lift amount and operating angle of the intake valve by the variable valve lift mechanism, the maximum lift amount and operation of the intake valve are controlled by driving the mechanism according to the engine operating state. There is a possibility that the vehicle is operated in a state where the angle is adjusted to a value different from the value when the feedback learning value is learned. Thus, in a situation where the engine is operated with the maximum lift amount and the operating angle of the intake valve adjusted to values different from the values when learning the feedback learning values, the feedback learning values are caused by the above causes. This is inappropriate as a value corresponding to a steady deviation of the air-fuel ratio from the stoichiometric air-fuel ratio. This is because when the maximum lift amount and operating angle of the intake valve change, the intake air amount of the internal combustion engine changes, and the ratio of the deviation amount to the appropriate value of the intake air amount of the engine due to the above causes to the total intake air amount is also This is because the magnitude of the influence of the actual air-fuel ratio on the steady-state deviation from the theoretical air-fuel ratio due to the above causes changes. For this reason, even if an attempt is made to compensate for the steady deviation by correcting the fuel injection amount using the feedback learning value, there is a possibility that the compensation cannot be performed accurately.

  In addition, in the internal combustion engine, when the maximum lift amount and the operating angle of the intake valve by the variable valve lift mechanism, such as during low load operation, cannot be reduced to the required amount only by varying the intake valve, the intake air amount can be reduced to the required amount. To reduce the amount, the throttle valve is driven from the fully open state to the closed side. Therefore, the internal combustion engine is operated in a state where the valve is driven so that the opening (intake pressure) differs from the opening of the throttle valve (corresponding to the intake pressure) when the learning of the feedback learning value is performed. There is a possibility that. In this way, in a situation where engine operation is performed with the throttle valve opening adjusted to a value different from the value obtained when learning the feedback learning value, the feedback learning value is the theory of the actual air-fuel ratio due to the above cause. It becomes inappropriate as a value corresponding to a steady deviation with respect to the air-fuel ratio. This is because when the throttle valve is driven from the fully open state to the closed side and the intake pressure of the internal combustion engine changes, the magnitude of the deviation from the appropriate value of the intake air amount of the engine due to the above cause depends on the intake pressure at that time. Instead, the magnitude of the influence on the steady deviation of the actual air-fuel ratio of the internal combustion engine from the stoichiometric air-fuel ratio changes accordingly. For this reason, even if it is attempted to compensate for the steady deviation by correcting the fuel injection amount using the feedback learning value, there is a possibility that the compensation cannot be performed accurately.

In order to deal with such a problem, it is conceivable to learn a feedback learning value as disclosed in Patent Document 1.
In this Patent Document 1, the feedback learning value is a first learning value corresponding to when the maximum lift amount and operating angle of the intake valve are small, and a second learning corresponding to when the maximum lift amount and operating angle of the intake valve is large. The value is calculated using an interpolation correction value variably set based on the current maximum lift amount and operating angle of the intake valve and the current intake pressure of the internal combustion engine.

  As the first learning value, for example, a value corresponding to the maximum lift amount and operating angle of the intake valve becomes a minimum value, and as the second learning value, for example, the maximum lift amount and operation of the intake valve are used. The value corresponding to the maximum angle is used. When the maximum lift amount and the operating angle of the intake valve are small (when the minimum value is found in Patent Document 1), the feedback learning value is updated by updating the first learning value based on the feedback correction value. The learning value is learned to obtain a value corresponding to a steady deviation of the actual air-fuel ratio of the internal combustion engine from the cause with respect to the theoretical air-fuel ratio. On the other hand, when the maximum lift amount and the operating angle of the intake valve are large (when the maximum value is reached in Patent Document 1), the feedback learning value is updated by updating the second learning value based on the feedback correction value. The learning value is learned to obtain a value corresponding to a steady deviation of the actual air-fuel ratio of the internal combustion engine from the cause with respect to the theoretical air-fuel ratio.

  In calculating the feedback learning value, in addition to the first learning value and the second learning value, interpolation correction that is variably set based on the current maximum lift amount and operating angle of the intake valve and the current intake pressure of the internal combustion engine A value is used. Therefore, the calculated feedback learning value can be a value corresponding to the maximum lift amount and operating angle of the intake valve and the intake pressure. Hereinafter, the calculation mode of the interpolation correction value will be described.

  As the maximum lift amount and operating angle of the intake valve increase, the interpolation correction value used for calculating the feedback learning value is reduced, and the sensitivity of the change in the feedback learning value with respect to the update of the first learning value or the second learning value is reduced. The This is because as the maximum lift amount and operating angle of the intake valve increase and the intake air amount increases, the ratio of the deviation amount to the appropriate value of the intake air amount due to the above cause decreases in the entire intake air amount. This is because the influence of the actual air-fuel ratio on the steady-state deviation from the stoichiometric air-fuel ratio becomes small, so that the feedback learning value is learned accordingly. Further, as the intake valve of the internal combustion engine decreases as the throttle valve is driven to the closed side, the interpolation correction value used for calculating the feedback learning value is reduced, and feedback for updating the first learning value and the second learning value is performed. The sensitivity of learning value changes is reduced. This is because, as the throttle valve is driven to the closed side and the intake pressure of the internal combustion engine decreases, the actual air-fuel ratio of the internal combustion engine with respect to the stoichiometric air-fuel ratio in the deviation from the appropriate value of the intake air amount of the engine due to the above causes This is because the influence on the shift is reduced and the feedback learning value is learned so as to correspond to the influence.

  By calculating the interpolation correction value as described above and calculating the feedback learning value using the interpolation correction value, the following effects can be obtained. In other words, the feedback learning value under the situation where the engine is operated in a state where the maximum lift amount and operating angle of the intake valve and the opening of the throttle valve are adjusted to values different from the values when learning the feedback learning value. Is suppressed from becoming inappropriate as a value corresponding to a steady shift of the actual air-fuel ratio to the stoichiometric air-fuel ratio due to the above cause. For this reason, even if it is intended to compensate for the steady deviation by correcting the fuel injection amount using the feedback learning value, it is possible to suppress a problem that the compensation cannot be performed accurately.

Next, a specific procedure for compensating for the steady deviation of the actual air-fuel ratio from the above cause with respect to the theoretical air-fuel ratio will be described.
The steady deviation of the actual air fuel ratio from the theoretical air fuel ratio due to the above causes appears when the maximum lift amount and operating angle of the intake valve are small. At this time, the first learning value is updated based on the feedback correction value that changes corresponding to the steady deviation, and the interpolation correction value based on the current maximum lift amount and operating angle of the intake valve and the current intake pressure. Is calculated. Then, by calculating the feedback learning value using the first learning value, the interpolation correction value, etc., learning to a value corresponding to the steady deviation of the actual air-fuel ratio from the actual air-fuel ratio due to the above cause of the feedback learning value. Is done. By correcting the fuel injection amount based on the learned feedback value in this way, compensation for the steady deviation can be achieved.

  However, when the maximum lift amount and operating angle of the intake valve are reduced and the throttle valve is driven from the fully open state to the closed side, the first learning value is updated based on the feedback correction amount and feedback learning is performed as described above. When learning of values is performed, the following problems may occur. That is, the learned feedback learning value may be a low-accuracy value as a value corresponding to a steady deviation of the actual air-fuel ratio of the internal combustion engine from the theoretical cause with respect to the theoretical air-fuel ratio.

  Here, when there is a steady deviation of the actual air fuel ratio from the theoretical air fuel ratio due to the above cause, the feedback learning value is a value corresponding to the steady deviation when the maximum lift amount and operating angle of the intake valve are small. In order to learn the same learning value with high accuracy, it is important to learn the feedback learning value with the throttle valve almost fully open. Therefore, as described above, the learning of the feedback learning value is performed by updating the first learning value in a state where the throttle valve is driven closer to the closing side than the fully opened state and the maximum lift amount and operating angle of the intake valve are small. Then, it becomes difficult to learn the learning value so that the feedback learning value becomes a value corresponding to the steady deviation.

  This is because when the maximum lift amount and operating angle of the intake valve are small, the throttle valve opening is adjusted to adjust the intake air amount of the internal combustion engine in the state where the throttle valve is driven to the closed side rather than fully open. This is related to the increased share of That is, when adjusting the intake air amount of the internal combustion engine, the share of the throttle valve opening adjustment for the adjustment increases, and the intake valve maximum lift amount for adjusting the intake air amount and The share by adjusting the operating angle is reduced. As described above, when the share by adjusting the maximum lift amount and the operating angle of the intake valve for adjusting the intake air amount of the internal combustion engine becomes small, the actual air-fuel ratio of the internal combustion engine due to the above cause becomes steady with respect to the stoichiometric air-fuel ratio. The effect on slack is reduced. For this reason, the first learning value corresponding to when the maximum lift amount and the operating angle of the intake valve are small is updated based on the feedback correction value so that the feedback learning value becomes a value corresponding to the steady deviation. Even if the feedback learning value is learned by updating the first learning value, it is difficult to make the learning value a highly accurate value.

  In order to suppress such a decrease in the accuracy of the feedback learning value, as shown in Patent Document 1, the throttle valve is driven from the fully opened state to the closed side, and the intake pressure of the internal combustion engine is lower than the intake pressure when the throttle valve is fully opened. It is conceivable that the learning of the feedback learning value is prohibited when the value falls below the determination value set in (1). By prohibiting learning of the feedback learning value in this way, the learning of the feedback learning value by updating the first learning value when the throttle valve is driven from the fully open state to the closing side causes A decrease in accuracy as a value corresponding to a steady deviation of the actual air-fuel ratio of the internal combustion engine from the stoichiometric air-fuel ratio is suppressed.

JP 2008-309119 A (paragraphs [0090] to [00102], [105] to [0107], FIGS. 10 to 13)

  As shown in Patent Document 1, by prohibiting learning of the feedback learning value when the intake pressure of the internal combustion engine is equal to or lower than the determination value, the theoretical air-fuel ratio of the actual air-fuel ratio of the internal combustion engine due to the above cause at the learning value It is possible to suppress a decrease in accuracy as a value corresponding to a steady deviation with respect to.

  However, if learning of the feedback learning value is prohibited when the intake pressure of the internal combustion engine becomes equal to or less than the determination value, the update of the first learning value corresponding to when the maximum lift amount and the operating angle of the intake valve are small, and the intake valve The update of the second learning value corresponding to the case where the maximum lift amount and the operating angle are large is not performed. Therefore, when the intake pressure of the internal combustion engine becomes equal to or lower than the determination value, not only learning of the feedback learning value by updating based on the feedback correction value of the first learning value is prohibited, but also by updating of the feedback correction value of the second learning value. Learning of the feedback learning value is also prohibited, and it is inevitable that the learning frequency of the learning value decreases. When the learning frequency of the feedback learning value decreases in this way, if there is a steady deviation of the actual air-fuel ratio from the theoretical air-fuel ratio due to causes other than the above causes, the feedback learning value by updating the second learning value It takes time until the feedback learning value is learned so as to be a value corresponding to the deviation, as much as learning of (2) is not performed.

  In addition, when the maximum lift amount and operating angle of the intake valve are fixed at the values during medium load operation, the adjustment of the intake air amount of the internal combustion engine during low load operation to the throttle valve closing side Since this is realized by driving, the intake pressure of the internal combustion engine becomes equal to or lower than the determination value, and learning of the feedback learning value is easily prohibited, and the learning frequency of the learning value is further reduced. As a result, the feedback learning value is used for correction of the fuel injection amount without learning so that the feedback learning value becomes a value corresponding to the steady deviation, and thereby the actual air-fuel ratio of the internal combustion engine is changed from the stoichiometric air-fuel ratio. There is a risk that effective exhaust purification by the three-way catalyst cannot be performed due to the deviation.

  The present invention has been made in view of such circumstances, and its purpose is to improve the learning accuracy by learning the feedback learning value by updating the first learning value when the throttle valve is driven from the fully opened state to the closed side. An object of the present invention is to provide an air-fuel ratio control device for an internal combustion engine that can suppress a decrease and suppress a decrease in the learning frequency of a feedback learning value.

In the following, means for achieving the above object and its effects are described.
In order to achieve the above object, according to the first aspect of the present invention, the valve lift variable that adjusts the intake air amount by varying the maximum lift amount and the operating angle of the intake valve that opens and closes to communicate and block the intake passage and the combustion chamber. A throttle valve that is provided in the intake passage and is normally opened while being normally opened while the amount of intake air cannot be reduced only by driving the variable valve lift mechanism; The present invention is applied to an internal combustion engine having a sensor provided in an exhaust passage and outputting a signal corresponding to the oxygen concentration in exhaust gas, and when the actual air-fuel ratio of the engine at the same output as the output of the sensor is a stoichiometric air-fuel ratio. A feedback correction value that increases or decreases based on a deviation from a target value that is a value, and a steady deviation of the actual air-fuel ratio from the theoretical air-fuel ratio based on the feedback correction value that increases or decreases. In the air-fuel ratio control apparatus for an internal combustion engine that controls the actual air-fuel ratio of the internal combustion engine to the stoichiometric air-fuel ratio by reflecting the feedback learning value learned so as to be a value to be applied to the fuel injection amount of the internal combustion engine, The feedback learning value includes a first learning value corresponding to the maximum lift amount and operating angle of the intake valve, a second learning value corresponding to a maximum lift amount and operating angle of the intake valve, Is calculated using an interpolation correction value variably set based on the maximum lift amount and operating angle of the intake valve and the current intake pressure of the internal combustion engine, and the interpolation correction value is calculated based on the interpolation correction value. The maximum lift of the intake valve is set so that the feedback learning value corresponds to the current maximum lift amount and operating angle of the intake valve and the current intake pressure. The feedback learning value is reduced as the operating angle increases and the intake pressure decreases, and the feedback learning value corresponds to a steady deviation of the actual air-fuel ratio from the stoichiometric air-fuel ratio. The learning of the feedback learning value to obtain a value is realized by updating the first learning value based on the feedback correction value when the interpolation correction value is greater than or equal to a predetermined determination value. When the value is less than the determination value, the second learning value is updated based on the feedback correction value.

  In an internal combustion engine, an assembly error around the intake valve and deposit accumulation on the valve cause a deviation from an appropriate value in the passage area of the connection portion of the intake passage with the combustion chamber. When the maximum lift amount and operating angle of the intake valve are reduced to reduce the intake air amount, a steady deviation from the stoichiometric air-fuel ratio occurs in the actual air-fuel ratio of the internal combustion engine. In order to suppress the steady deviation of the actual air-fuel ratio from the theoretical air-fuel ratio, learning of the learning value is performed so that the feedback learning value becomes a value corresponding to the steady deviation, and the feedback learning value after learning becomes This is reflected in the fuel injection amount of the internal combustion engine. Regarding learning of the feedback learning value, specifically, when the maximum lift amount and operating angle of the intake valve are small, the maximum lift amount of the intake valve and the intake valve are determined based on the feedback correction value that changes corresponding to the steady deviation. This is realized by updating the first learning value corresponding to the small operating angle.

  By the way, in order to reduce the intake air amount of the internal combustion engine to the required amount at the time of low load operation of the internal combustion engine or when the maximum lift amount of the intake valve and the operating angle are fixed at the medium load operation value, the throttle valve Is driven from the fully open state to the closed side value. When the throttle valve is driven to the closing side in this way, the intake pressure of the internal combustion engine decreases, and the interpolation correction value used for calculation of the feedback learning value is accordingly reduced. The interpolation correction value is set to a smaller value as the maximum lift amount and operating angle of the intake valve are increased through driving of the variable valve lift mechanism. According to the above configuration, the learning of the feedback learning value is performed in a different manner depending on the interpolation correction value variably set based on the maximum lift amount and operating angle of the intake valve, the intake pressure of the internal combustion engine, and the like.

  That is, only when the interpolation correction value is greater than or equal to the determination value, the first learning value corresponding to when the maximum lift amount and operating angle of the intake valve is small is updated based on the feedback correction value. The learning value is learned to obtain a value corresponding to a steady deviation of the actual air-fuel ratio from the stoichiometric air-fuel ratio. Note that the interpolation correction value becomes larger than the determination value because the throttle valve is fully opened or almost fully opened, the intake pressure of the internal combustion engine is increased, and the maximum lift amount and operating angle of the intake valve are decreased, thereby This is when the interpolation correction value increases. Therefore, when the maximum lift amount and the operating angle of the intake valve are small, and when the throttle valve is driven to the closed side rather than the fully opened state, in other words, when the interpolation correction value is less than the determination value, the first learning is performed. The feedback learning value is not learned by updating the value. And, by learning such a feedback learning value, the accuracy of the feedback learning value as a value corresponding to a steady deviation of the actual air-fuel ratio of the internal combustion engine from the above cause with respect to the theoretical air-fuel ratio is reduced. It is suppressed.

  On the other hand, when the interpolation correction value is less than the determination value, the second learning value corresponding to the maximum lift amount and operating angle of the intake valve is updated based on the feedback correction value, whereby the feedback learning value is changed to the actual air-fuel ratio. Learning of the learning value is executed to obtain a value corresponding to a steady deviation with respect to the theoretical air-fuel ratio. Note that the interpolation correction value is less than the determination value when the maximum lift amount and the operating angle of the intake valve are increased, thereby reducing the interpolation correction value. Even if the maximum lift amount and operating angle of the intake valve are small, the interpolation correction value may be less than the judgment value when the throttle valve is driven to the closed side rather than the fully opened state and the intake pressure of the internal combustion engine becomes low. Is expensive. Therefore, when the maximum lift amount and operating angle of the intake valve are small and the throttle valve is driven to the closed side rather than the fully opened state, it corresponds to the case where the maximum lift amount and operating angle of the intake valve are large. It is possible to learn the feedback learning value by updating the second learning value. For this reason, in the state as described above, learning of the feedback learning value by updating the first learning value and learning of the feedback learning value by updating the second learning value are both prohibited, and the learning frequency of the learning value is thereby reduced. Decreasing is suppressed. Specifically, when there is a steady deviation of the actual air-fuel ratio of the internal combustion engine from the cause other than the above-described cause to the stoichiometric air-fuel ratio, the feedback is made by the amount that the second learning value cannot be updated due to the prohibition of learning. It is suppressed that the learning frequency of a learning value falls. As a result, there is no problem that it takes time until the feedback learning value is learned so as to be a value corresponding to the steady deviation.

  As described above, it is possible to suppress the learning accuracy of the learning value from being lowered by learning the sub feedback learning value based on the update of the first learning value in the state where the throttle valve is driven from the fully open state to the closed side, and The learning of the sub-feedback learning value based on the update of the second learning value in the state can suppress the decrease in the learning frequency of the learning value.

  According to a second aspect of the present invention, in the first aspect of the invention, the first learning value is a value corresponding to a maximum lift amount and an operating angle of the intake valve that are minimum values, and the second learning value. The value is a value corresponding to the maximum lift amount and operating angle of the intake valve reaching a maximum value, and the interpolation correction value is based on the maximum lift amount and operating angle of the intake valve and the intake pressure. It is variable in the range of “0” to “1.0”, and decreases toward “0” as the maximum lift amount and operating angle increase, and decreases toward “0” as the intake pressure decreases. The feedback learning value is “SG”, the first learning value is “A”, the second learning value is “B”, and the interpolation correction value is “K”. Then, the following formula “SG = B + (A− ) · K ”, and the determination value is the center or near the center of the frequently used range of the variable correction range from“ 0 ”to“ 1.0 ”in the interpolation correction value. The gist is that the value of

  According to the above configuration, since the determination value is set to a value in the center or near the center of the frequently used range of the variable setting range from “0” to “1.0” in the interpolation correction value, the interpolation correction is performed. There is a high possibility that the opportunity for the value to be greater than or equal to the determination value and the opportunity for the value to be less than the determination value occur equally. Therefore, the frequency at which learning of the feedback learning value is performed by updating the first learning value when the interpolation correction value is greater than or equal to the determination value, and the feedback learning value by updating the second learning value when the interpolation correction value is less than the determination value Can be equalized with the frequency of learning.

  According to a third aspect of the present invention, in the first or second aspect of the present invention, the internal combustion engine has a plurality of cylinders, and the sensor includes a collection portion of exhaust passages connected to the cylinders, and more than that. A sensor provided between the catalyst on the downstream side and a sensor provided on the downstream side of the catalyst, and the feedback correction value is an output between the sensor output downstream of the catalyst and the target value. It is increased or decreased based on the deviation, and is calculated as a fuel injection amount correction value for making the actual air-fuel ratio of the internal combustion engine the stoichiometric air-fuel ratio based on the output of the sensor upstream of the catalyst, and is used for correcting the fuel injection amount A sub-feedback correction value reflected in the main feedback correction value, and the feedback learning value is based on the sub-feedback correction value. It was summarized in that a sub-feedback learning value is learned to be a value corresponding to the stationary difference is reflected in the main feedback correction value for.

  According to the above configuration, the actual air-fuel ratio in the internal combustion engine is controlled to the stoichiometric air-fuel ratio by correcting the fuel injection amount with the main feedback correction value. The main feedback correction value is calculated based on the output of a sensor provided on the upstream side of the catalyst and in the vicinity of the collective portion of the exhaust passage connected to each cylinder. In this way, it is inevitable that the way the exhaust gas sent out from each cylinder in order with respect to the sensor provided in this way differs for each exhaust gas from each cylinder. Further, the oxygen concentration contained in the exhaust gas from each cylinder varies depending on the combustion state of the fuel in each cylinder. For this reason, the output of the sensor largely fluctuates depending on whether the exhaust corresponding to the sensor is exhausted from any of the cylinders, and the main feedback correction value fluctuates frequently and greatly accordingly.

  Here, assuming that a steady deviation between the actual air-fuel ratio and the stoichiometric air-fuel ratio in an internal combustion engine is to be learned as a learning value based on the main feedback correction value, the main feedback correction value is as described above. Since it is frequently and greatly fluctuates for a reason, the learned value is not always highly accurate as a value corresponding to the steady deviation. From this, it is possible to suppress frequent and large fluctuations in the main feedback correction value due to the large fluctuations for the above reason in the output of the sensor upstream from the catalyst, and through correction of the fuel injection amount by the main feedback correction value. The sub-feedback correction value that increases or decreases based on the output of the sensor downstream of the catalyst, and the sub-feedback correction value with respect to the main feedback correction value for the purpose of removing the influence of the air-fuel ratio control accuracy of the catalyst, etc. The sub-feedback learning value learned based on is reflected.

  Here, regarding the output of the sensor on the downstream side of the catalyst, since the sensor is located on the downstream side with respect to the collective portion in the exhaust passage, the contact of the exhaust sent out sequentially from each cylinder to the sensor. This is unlikely to be different for each exhaust from each cylinder, and is not easily affected by the difference in how the exhaust hits. Accordingly, the sub feedback correction value that increases or decreases based on the output from the sensor and the sub feedback learning value that is learned based on the sub feedback correction value are reflected in the main feedback correction value, so that the main feedback correction value described above is frequently used. In addition, large fluctuations can be suppressed. In addition, since the output of the sensor becomes a value corresponding to the oxygen concentration in the exhaust gas after passing through the catalyst, the sub feedback correction value and the sub feedback learning value are reflected in the main feedback correction value, thereby correcting the main feedback correction. It is also possible to remove the influence of the catalyst on the air-fuel ratio control accuracy through the correction of the fuel injection amount by the value.

  According to the above configuration, an effect equivalent to that of the first aspect of the invention can be obtained with respect to learning of such a sub-feedback learning value. Accordingly, the fuel injection amount is corrected by the main feedback correction value reflecting the sub-feedback learning value, thereby compensating for the steady deviation between the actual air-fuel ratio and the stoichiometric air-fuel ratio of the internal combustion engine. Can be performed accurately.

1 is a schematic diagram showing an entire engine to which an air-fuel ratio control apparatus of an embodiment is applied. 1 is a schematic diagram showing a mounting position of an air-fuel ratio sensor in an exhaust system of the engine. The timing chart which shows the change aspect of the lift amount of the intake valve with respect to the change of a crank angle. (A) And (b) is a graph which shows the ratio of the drive part of a throttle valve, and the drive part of a valve lift variable mechanism in adjusting the intake air amount of an engine to a required amount. The graph which shows the difference for every cylinder of the way of exhaust_gas | exhaustion with respect to an air fuel ratio sensor, and the difference for every cylinder of the air fuel ratio of air-fuel | gaseous mixture. The graph which shows transition of the interpolation correction value with respect to the change of the maximum lift amount and operating angle of an intake valve, and the change of the intake pressure of an engine. The graph which shows the relationship between the change of the maximum lift amount and operating angle of an intake valve, and the passage area of the communication part of the intake passage with the combustion chamber. The graph which shows the relationship between the change of the maximum lift amount and the operating angle of an intake valve, and the steady deviation with respect to the theoretical air fuel ratio of an actual air fuel ratio. The graph which shows transition of the sub feedback learning value with respect to the change of the maximum lift amount of an intake valve, an operating angle, and the change of the intake pressure of an engine. The flowchart which shows the procedure which determines whether the learning is implement | achieved by the update of a 1st learning value, or the update of a 2nd learning value in learning of a subfeedback learning value. The graph which shows transition of the interpolation correction value with respect to the change of the maximum lift amount and operating angle of an intake valve, and the change of the intake pressure of an engine.

Hereinafter, an embodiment in which the present invention is embodied in an air-fuel ratio control apparatus for an automobile four-cylinder engine will be described with reference to FIGS.
In the engine 1 shown in FIG. 1, a throttle valve 13 is provided in an intake passage 3 connected to the combustion chamber 2 so as to be openable and closable. Air is sucked into the combustion chamber 2 through the intake passage 3 and fuel injection is performed. The fuel injected from the valve 4 is supplied to the combustion chamber 2. When the air / fuel mixture is ignited by the spark plug 5, the air / fuel mixture burns, the piston 6 reciprocates, and the crankshaft 7 that is the output shaft of the engine 1 rotates.

  On the other hand, the air-fuel mixture after combustion in the combustion chamber 2 is sent as exhaust gas from each combustion chamber 2 to the exhaust passage 8 and purified by the three-way catalyst of the catalytic converter 16 provided in the exhaust passage 8. Released to the outside. This three-way catalyst can most effectively remove all harmful components (HC, CO, NOx) in the exhaust when the oxygen concentration in the catalyst atmosphere becomes the value at the time of combustion of the air-fuel mixture at the stoichiometric air-fuel ratio It is.

  In the exhaust passage 8, air-fuel ratio sensors 17 and 18 are provided on the upstream side and downstream side of the catalytic converter 16 for outputting linear detection signals corresponding to the oxygen concentration in the exhaust gas. More specifically, as shown in FIG. 2, the air-fuel ratio sensor 17 is located in the vicinity of the collective portion of the exhaust passage 8 connected to the combustion chambers 2 of the four cylinders # 1, # 2, # 3, and # 4 in the engine 1, respectively. Thus, the air-fuel ratio sensor 18 is provided on the downstream side of the catalytic converter 16.

  In the engine 1 shown in FIG. 1, communication between the combustion chamber 2 and the intake passage 3 is made open / closed by the opening / closing operation of the intake valve 9, and between the combustion chamber 2 and the exhaust passage 8 is made by opening / closing operation of the exhaust valve 10. Communication / blocking. The intake valve 9 and the exhaust valve 10 are opened and closed with the rotation of the intake camshaft 11 and the exhaust camshaft 12 to which the rotation of the crankshaft 7 is transmitted. Further, the engine 1 is provided with a variable valve lift mechanism 14 that makes the maximum lift amount and operating angle of the intake valve 9 variable. The variable valve lift mechanism 14 changes the maximum lift amount and the operating angle of the intake valve 9 in synchronization with each other as shown in FIG. 3 through rotational driving of the electric motor 15 within a predetermined rotation angle range. .

Next, the electrical configuration of the air-fuel ratio control apparatus in the present embodiment will be described with reference to FIG.
The air-fuel ratio control device includes an electronic control device 21 that executes various controls relating to the engine 1. The electronic control unit 21 is a CPU that executes various arithmetic processes related to the above control, a ROM that stores programs and data necessary for the control, a RAM that temporarily stores the arithmetic results of the CPU, etc. The input / output port for inputting / outputting is provided.

In addition to the air-fuel ratio sensors 17 and 18, the following various sensors are connected to the input port of the electronic control unit 21.
An accelerator position sensor 28 that detects the amount of depression (accelerator depression amount) of the accelerator pedal 27 that is depressed by the driver of the automobile.

A throttle position sensor 30 that detects the opening (throttle opening) of the throttle valve 13 provided in the intake passage 3.
An air flow meter 32 for detecting the amount of air taken into the combustion chamber 2 through the intake passage 3;

An intake pressure sensor 33 that detects a pressure (intake pressure) downstream of the throttle valve 13 in the intake passage 3.
A crank position sensor 34 that outputs a signal corresponding to the rotation of the crankshaft 7 and is used for calculation of the engine rotation speed or the like.

A position sensor 35 for detecting a rotation angle that is a value within the predetermined rotation angle range of the electric motor 15.
The output port of the electronic control device 21 is connected to the fuel injection valve 4, the throttle valve 13, the drive circuit for the electric motor 15, and the like.

  The electronic control unit 21 grasps the engine operating state such as the engine speed and the engine load (the amount of air taken into the combustion chamber 2 per cycle of the engine 1) based on the detection signals input from the various sensors. To do. The engine speed is obtained based on a detection signal from the crank position sensor 34. The engine load is calculated from the intake air amount of the engine 1 obtained based on detection signals from the accelerator position sensor 28, the throttle position sensor 30, the air flow meter 32, and the like, and the engine rotation speed. The electronic control unit 21 outputs command signals to various drive circuits connected to the output port according to the engine operating state such as the engine load and the engine speed. Thus, fuel injection amount control, intake air amount control, and the like in the engine 1 are performed through the electronic control unit 21.

  The intake air amount control of the engine 1 is performed by driving the variable valve lift mechanism 14 and opening / closing the throttle valve 13. Specifically, for the purpose of keeping the pumping loss of the engine 1 low, the throttle valve 13 is fully opened as much as possible, and only the maximum lift amount and the operating angle of the intake valve 9 by the drive of the valve lift variable mechanism 14 are variable. The intake air amount of the engine 1 is adjusted to the required amount.

  When the intake air amount is controlled to the required amount only by driving the variable valve lift mechanism 14 while the throttle valve 13 is fully opened, the drive amount of the variable valve lift mechanism 14 and the drive of the throttle valve 13 to realize this control. The ratio with the minute is a ratio as shown in FIG. That is, the driving amount of the variable valve lift mechanism 14 is “100%”, and the driving amount of the throttle valve 13 is “0%”. By adjusting the intake air amount while the throttle valve 13 is fully opened in this way, the intake pressure can be increased to a value close to atmospheric pressure, and the pumping loss of the engine 1 can be kept low. .

  However, when the engine 1 is in an operating state where the required amount of intake air amount is small, such as during low load operation including idle operation, the variable of the maximum lift amount and operating angle of the intake valve 9 by driving the variable valve lift mechanism 14 is not sufficient. The intake air volume cannot be reduced to the required level. Therefore, under such an engine operating state, in addition to driving the variable valve lift mechanism 14, the throttle valve 13 is also driven from the fully open to the closed side, and the intake air amount in the engine 1 is reduced to the required amount. Is planned. Further, when the maximum lift amount and the operating angle of the intake valve 9 cannot be reduced due to the sticking abnormality of the variable valve lift mechanism 14 and the intake air amount cannot be reduced to the required amount, the throttle valve 13 can be closed from the fully opened side to the closed side. The intake air amount is reduced to the required amount by driving to.

  When the intake air amount is controlled to the required amount in this way, the ratio between the drive amount of the variable valve lift mechanism 14 and the drive amount of the throttle valve 13 for realizing this is, for example, (b) in FIG. The ratio is as shown in. The drive amount of the variable valve lift mechanism 14 becomes smaller than “100%” as the required amount of intake air decreases, and the drive amount of the throttle valve 13 decreases to “0%” as the required amount decreases. Will be larger than The increase in the amount of drive of the throttle valve 13 in this way means that the throttle valve 13 is driven closer to the closed side, and the intake pressure of the engine 1 decreases through the drive of the valve 13. To do.

  The fuel injection amount control of the engine 1 is based on the engine rotational speed, the engine load, etc., and calculates the fuel injection amount required at that time as the command injection amount Q, and the amount of fuel corresponding to the command injection amount Q is calculated. This is realized by driving the fuel injection valve 4 to be injected. The command injection amount Q used in such fuel injection amount control is calculated using the following equation (1) based on the basic fuel injection amount Qbase, the main feedback correction value DF, and the main feedback learning value MG (i). The

Q = Qbase + DF + MG (i) (1)
Q: Instructed injection amount
Qbase: Basic fuel injection amount
DF: Main feedback correction value
MG (i): main feedback learning value Here, the basic fuel injection amount Qbase is a theoretical fuel injection amount necessary for obtaining a mixture of the stoichiometric air-fuel ratio, and is a detection signal from the air flow meter 32 or the like. Based on the intake air amount GA of the engine 1 and the theoretical air-fuel ratio “14.7” obtained based on this, this is a value calculated using the equation “Qbase = GA / 14.7 (3)”.

  The main feedback correction value DF is for correcting the fuel injection amount (basic fuel injection amount Qbase), and is detected by the air-fuel ratio sensor 17 so that the actual air-fuel ratio of the engine 1 becomes the stoichiometric air-fuel ratio (target value). It is increased or decreased based on the actual air-fuel ratio of the engine 1 obtained from the signal. Through such increase / decrease of the main feedback correction value DF, the command injection amount Q is increased / decreased so that the actual air / fuel ratio of the engine 1 becomes the stoichiometric air / fuel ratio, thereby realizing main feedback control for setting the actual air / fuel ratio to the stoichiometric air / fuel ratio. Is done.

  The main feedback learning value MG (i) is used to correct the fuel injection amount (basic fuel injection amount Qbase) in the same way as the main feedback correction value DF, and is used for clogging the intake system and the fuel injection system in the engine 1. The engine 1 is updated so as to have a value that compensates for the steady deviation of the air-fuel ratio of the engine 1 from the stoichiometric air-fuel ratio. The update of the main feedback learning value MG (i) is performed based on the main feedback correction value DF. Then, through the correction of the fuel injection amount by the main feedback learning value MG (i) and the main feedback correction value DF, and the update of the main feedback learning value MG (i), the learning value MG (i) is changed to the above steady state. The main feedback learning control that learns as a value corresponding to the deviation is realized.

  Next, the calculation procedure of the main feedback correction value DF in the main feedback control and the update procedure of the main feedback learning value MG (i) in the main feedback learning control will be described individually.

[Calculation of main feedback correction value DF]
The main feedback correction value DF is calculated using the following equation (4) based on the fuel amount deviation ΔQ, the proportional gain Gp, the fuel amount deviation integrated value ΣΔQ, and the integral gain Gi.

DF = ΔQ · Gp + ΣΔQ · Gi (4)
DF: Main feedback correction value
ΔQ: Fuel amount deviation
Gp: Proportional gain (negative value)
ΣΔQ: Fuel amount deviation integrated value
Gi: integral gain (negative value)
In equation (4), the term “ΔQ · Gp” on the right side is a proportional term that takes a magnitude proportional to the amount of deviation of the actual air-fuel ratio from the stoichiometric air-fuel ratio, and fuel corresponding to the amount of deviation. This is to increase or decrease the injection amount to bring the actual air-fuel ratio closer to the theoretical air-fuel ratio.

  The fuel amount deviation ΔQ used in the proportional term “ΔQ · Gp” is a value obtained by subtracting the theoretical fuel amount necessary to obtain the stoichiometric air-fuel ratio mixture from the actually burned fuel amount, Based on the intake air amount GA, the actual air-fuel ratio ABF, and the basic fuel injection amount Qbase, it is calculated using the equation “ΔQ = (GA / ABF) −Qbase (5)”. The actual air-fuel ratio ABF is calculated using the expression “ABF = g (VAF) (6)” based on the output VAF of the air-fuel ratio sensor 17.

  The output VAF of the air-fuel ratio sensor 17 decreases as the oxygen concentration in the exhaust gas upstream of the catalyst decreases. When the air-fuel mixture is burned at the stoichiometric air-fuel ratio, the output VAF corresponds to the oxygen concentration in the exhaust gas at that time. For example, “0v”. Therefore, the output VAF of the air-fuel ratio sensor 17 becomes smaller than “0 v” as the oxygen concentration in the exhaust gas upstream of the catalyst is reduced due to the combustion of the rich air-fuel mixture (rich combustion). Further, as the oxygen concentration in the exhaust gas upstream of the catalyst increases due to the combustion of the lean air-fuel mixture (lean combustion), the value becomes larger than the output VAF “0 v” of the air-fuel ratio sensor 17.

The proportional gain Gp used in the proportional term “ΔQ · Gp” is a constant obtained in advance by experiments or the like, and is set as a negative value here.
In the equation (4), the term “ΣΔQ · Gi” on the right side represents the residual deviation between the actual air-fuel ratio and the stoichiometric air-fuel ratio that cannot be canceled out only by increasing or decreasing the fuel injection amount by the proportional term “ΔQ · Gp”. This is an integral term for eliminating the difference between the actual air-fuel ratio and the stoichiometric air-fuel ratio by increasing or decreasing the fuel injection amount corresponding to the residual deviation.

  The fuel amount deviation integrated value ΣΔQ used in the integral term “ΣΔQ · Gi” is a value obtained through integration processing in which the fuel amount deviation ΔQ is added at predetermined time intervals. In this integration process, an equation “ΣΔQ ← previous ΣΔQ + ΔQ (7)” is executed at predetermined time intervals. Further, the integral gain Gi used in the integral term “ΣΔQ · Gi” is a constant obtained in advance by experiments or the like, and is set as a negative value here.

  Accordingly, when the actual amount of fuel burned is too small and the actual air-fuel ratio ABF becomes large (lean), the fuel amount deviation ΔQ calculated by the above equation (5) changes in the negative direction. The main feedback correction value DF calculated by equation (4) increases. On the other hand, when the actual amount of fuel burned is too large and the actual air-fuel ratio ABF becomes small (rich), the fuel amount deviation ΔQ changes in the positive direction. DF decreases.

  As described above, by increasing or decreasing the main feedback correction value DF based on the actual air-fuel ratio ABF, the command injection amount Q increases or decreases, and the fuel injection amount of the engine 1 becomes the stoichiometric air-fuel ratio. Adjusted.

[Update of main feedback learning value MG (i)]
The main feedback learning value MG (i) is when the feedback correction rate, which is the ratio of the main feedback correction value DF to the basic fuel injection amount Qbase, is 1% or more, for example, and the main feedback correction value DF is stable Updated to Specifically, based on the expression “MG (i) ← latest DF (8)”, the main feedback correction value DF at that time is set as the main feedback learning value MG (i), so that the learning value MG (i) is updated.

  Therefore, when the main feedback correction value DF is large, the main feedback learning value MG (i) is updated to the increasing side, and the engine 1 is corrected through correction to the increasing side of the command injection amount Q by the learning value MG (i). The amount of fuel injection is increased. When the main feedback correction value DF is small, the main feedback learning value MG (i) is updated to the decreasing side, and the engine 1 is corrected through the correction to the decreasing side of the command injection amount Q by the learning value MG (i). The fuel injection amount is reduced.

  Through the update of the main feedback learning value MG (i) and the correction of the fuel injection amount based on the learning value MG (i) as described above, the main feedback correction value DF comes close to “0”. Further, the main feedback learning value MG (i) when the main feedback correction value DF is stabilized to approach “0” to some extent is the stoichiometric air-fuel ratio of the engine 1 caused by clogging of the intake system or the fuel injection system. A value corresponding to a steady deviation with respect to. Then, the main feedback learning value MG (i) at this time becomes a value learned as a value corresponding to the steady deviation.

  The main feedback learning value MG (i) is prepared for each of a plurality of learning regions i (i = 1, 2, 3,...) Divided according to the engine load region. When the learning region i corresponding to the operating state changes according to the change in the operating state of the engine 1, the updated main feedback learning value MG (i) also corresponds to the learning region i after the change. And can be switched. Thus, the main feedback learning value MG (i) is updated for each learning region i.

  However, although the main feedback learning value MG (i) learned as described above can ensure a certain degree of accuracy as a value corresponding to a steady deviation of the air-fuel ratio in the engine 1 from the theoretical air-fuel ratio, The value is not so high. This is calculated based on the output VAF of the air-fuel ratio sensor 17 provided in the vicinity of the collection portion of the exhaust passage 8 connected to each of the cylinders # 1 to # 4 on the upstream side of the main feedback correction value DF. It is related to being. That is, the manner in which the exhaust sent out sequentially from the cylinders # 1 to # 4 with respect to the air-fuel ratio sensor 17 provided in this way is, for example, as shown in FIG. 5 for each exhaust of each cylinder # 1 to # 4. It is inevitable that it will be different. The oxygen concentration in the exhaust gas from each cylinder # 1 to # 4, in other words, the air-fuel ratio in the air-fuel mixture in each cylinder # 1 to # 4 is the combustion state of the fuel in each cylinder # 1 to # 4. For example, as shown in FIG. For this reason, the output VAF of the air-fuel ratio sensor 17 fluctuates greatly depending on whether the exhaust that hits the sensor 17 is exhausted from each of the cylinders # 1 to # 4, and the main feedback correction value DF is also adjusted accordingly. Fluctuates frequently and greatly. Since the main feedback learning value MG (i) is learned based on the main feedback correction value DF that frequently and greatly fluctuates, the learning value MG (i) is accurate as a value corresponding to the steady deviation. It ’s only not so high.

  In this embodiment, the main feedback correction value DF is frequently and greatly fluctuated due to variations in output characteristics or changes with time in the above-described air-fuel ratio sensor 17, and the main feedback control accuracy is thereby prevented from degrading. Therefore, the sub feedback control is performed. In addition, sub-feedback learning control is also performed in order to more accurately compensate a steady deviation of the air-fuel ratio of the engine 1 from the stoichiometric air-fuel ratio caused by the air-fuel ratio sensor 17 or the catalyst. Hereinafter, sub feedback control and sub feedback learning control will be described in detail.

  In the sub feedback control and the sub feedback learning control, the main feedback correction value DF is corrected by the sub feedback correction value VH and the sub feedback learning value SG, and thereby the sub feedback correction value VH and the sub feedback are added to the correction value DF. The learning value SG is reflected. Specifically, the output VAF of the air-fuel ratio sensor 17 is corrected by the sub-feedback correction value VH and the sub-feedback learning value SG based on the following formula (9), and the main feedback is calculated based on the formula (4) using the corrected output VAF. By calculating the correction value DF, the correction value DF is corrected by the correction value VH and the learning value SG.

VAF ← Latest VAF + VH + SG (9)
VAF: Output of air-fuel ratio sensor
VH: Sub feedback correction value
SG: Sub-feedback learning value The sub-feedback correction value VH is increased or decreased according to the detection signal from the air-fuel ratio sensor 18 downstream of the catalyst. Through the correction of the main feedback correction value DF by the sub feedback correction value VH that increases or decreases in this way, the command injection amount Q is increased or decreased, thereby realizing the sub feedback control that suppresses the decrease in the accuracy of the main feedback control. By executing the sub feedback control, the sub feedback correction value VH changes to a value that suppresses the decrease in accuracy of the main feedback control.

  The sub-feedback learning value SG is updated (learned) based on the sub-feedback correction value VH so as to compensate for a steady deviation of the air-fuel ratio of the engine 1 from the theoretical air-fuel ratio. Through the correction of the main feedback correction value DF by the sub feedback correction value VH and the sub feedback learning value SG, and the update (learning) of the sub feedback learning value SG, the engine 1 emptying caused by the air-fuel ratio sensor 17, the catalyst, or the like. Sub-feedback learning control that compensates for a steady deviation of the fuel ratio from the stoichiometric air-fuel ratio is realized.

  Here, the output of the air-fuel ratio sensor 18 is sent in order from each cylinder # 1 to # 4 with respect to the sensor 18 because the sensor 18 is located on the downstream side with respect to the collective portion in the exhaust passage 8. It is difficult for the way of exhausted exhaust to be different for each exhaust from each cylinder # 1 to # 4, and it is difficult to be affected by the difference in the manner of exhausting. Therefore, the sub feedback correction value VH that increases or decreases based on the output from the sensor 18 and the sub feedback learning value SG that is learned based on the sub feedback correction value VH are reflected in the main feedback correction value DF, thereby correcting the correction value. It becomes possible to suppress the above-mentioned frequent and large fluctuation of DF. Further, since the output of the air-fuel ratio sensor 18 becomes a value corresponding to the oxygen concentration in the exhaust gas after passing through the catalyst, the sub feedback correction value VH and the sub feedback learning value SG are reflected in the main feedback correction value DF. Thus, it is possible to remove the influence of catalyst deterioration or the like on the accuracy of the air-fuel ratio control through the correction of the fuel injection amount by the correction value DF.

  Further, the output of the air-fuel ratio sensor 18 is not easily affected by the difference in how the exhaust gas sent from each cylinder in turn hits the sensor 18. Therefore, the sub-feedback learning value SG is more accurate as a value corresponding to a steady deviation between the actual air-fuel ratio of the engine 1 and the stoichiometric air-fuel ratio than the main feedback learning value MG (i). Become. By reflecting this sub-feedback learning value SG on the main feedback learning value MG (i), it is possible to more accurately compensate for a steady deviation of the air-fuel ratio of the engine 1 from the stoichiometric air-fuel ratio.

  Next, the calculation procedure of the sub feedback correction value VH in the sub feedback control and the calculation procedure of the sub feedback learning value SG in the sub feedback learning control will be individually described.

[Calculation procedure of sub feedback correction value VH]
The sub feedback correction value VH is based on the output VO of the air-fuel ratio sensor 18, the proportional gain Kp, the output integrated value ΣVO, the integral gain Ki, the voltage differential value dV, and the differential gain Kd using the following equation (10). Calculated.

VH = VO · Kp + ΣVO · Ki + dV · Kd (10)
VH: Sub feedback correction value
VO: Air-fuel ratio sensor output
Kp: Proportional gain (positive value)
ΣVO: Output integrated value
Ki: integral gain (positive value)
dV: Voltage differential value
Kd: differential gain (positive value)
In equation (10), the term “VO · Kp” on the right side is a magnitude proportional to the amount of deviation between the actual value of the oxygen concentration downstream of the catalyst and the value when combustion at the stoichiometric air-fuel ratio is performed. The main feedback correction value DF (output VAF) is increased or decreased by an amount corresponding to the deviation amount, and the deviation amount is brought close to “0”.

  The output VO of the air-fuel ratio sensor 18 used in this proportional term “VO · Kp” becomes smaller as the oxygen concentration in the exhaust gas downstream of the catalyst becomes thinner. When the air-fuel mixture is burned at the stoichiometric air-fuel ratio, Corresponding to the oxygen concentration in the exhaust at that time, for example, “0 v”. Accordingly, when the oxygen concentration in the exhaust gas downstream of the catalyst becomes higher than the value when the air-fuel mixture is burned at the stoichiometric air-fuel ratio due to the lean combustion, etc., the air-fuel ratio sensor 18 determines from “0 v”. A larger value is output. Further, if the oxygen concentration in the exhaust gas downstream of the catalyst becomes thinner than the value when the air-fuel mixture of the stoichiometric air-fuel ratio is burned due to the rich combustion or the like, the air-fuel ratio sensor 18 determines from “0 v”. A smaller value is output. The proportional gain Kp used in the proportional term “VO · Kp” is a constant obtained in advance through experiments or the like, and is set as a positive value here.

  In the equation (10), the term “ΣVO · Ki” on the right side is a residual deviation that cannot be canceled only by the increase / decrease in the main feedback correction value DF (output VAF) by the proportional term “VO · Kp”, that is, downstream of the catalyst. This is an integral term for eliminating the residual deviation between the actual value of the oxygen concentration and the value when combustion at the stoichiometric air-fuel ratio is performed. The integral term “ΣΣVO · Ki” becomes a value corresponding to the residual deviation, and the main feedback correction value DF (output VAF) is increased or decreased by the amount corresponding to the integral term “ΣVO · Ki”. The actual value for the oxygen concentration of the gas and the value obtained when combustion at the stoichiometric air-fuel ratio is performed can be achieved.

  The output integrated value ΣVO used in the integral term “ΣVO · Ki” is a value obtained through integration processing in which the output VO of the air-fuel ratio sensor 17 is added at predetermined time intervals. In this integration process, an equation “ΣVO ← previous ΣVO + VO (12)” is executed at predetermined time intervals. The integral gain Ki used in the integral term “ΣVO · Ki” is a constant obtained in advance through experiments or the like, and is set here as a positive value.

  In the expression (10), the term “dV · Kd” on the right side indicates the deviation amount between the actual value of the oxygen concentration downstream of the catalyst and the value when combustion at the stoichiometric air-fuel ratio is performed with good responsiveness. This is a differential term for convergence to “0”.

  The voltage differential value dV used for the differential term “dV · Kd” is a value obtained by time-differentiating the output VO of the air-fuel ratio sensor 18 and represents the amount of change per unit time of the output VO. . The differential gain Kd used for the differential term “dV · Kd” is a constant obtained in advance by experiments or the like, and is set here as a positive value.

  Therefore, when the oxygen concentration in the exhaust gas downstream of the catalyst becomes thinner than the value at the time of combustion at the stoichiometric air-fuel ratio (during rich combustion), the output VO of the air-fuel ratio sensor 18 changes in the negative direction. The sub feedback correction value VH calculated by the equation (10) decreases. In other words, the output VO of the air-fuel ratio sensor 18 changes in the negative direction based on the deviation between the actual air-fuel ratio and the theoretical air-fuel ratio, and the sub feedback correction value VH decreases. On the contrary, when the oxygen concentration in the exhaust gas downstream of the catalyst is higher than the value at the time of combustion at the stoichiometric air-fuel ratio (during lean combustion), the output VO of the air-fuel ratio sensor 18 changes in the positive direction. Therefore, the sub feedback correction value VH increases. In other words, the output VO of the air-fuel ratio sensor 18 changes in the positive direction based on the deviation between the actual air-fuel ratio and the theoretical air-fuel ratio, and the sub feedback correction value VH increases.

  As described above, the sub-feedback correction value VH is increased or decreased based on the oxygen concentration in the exhaust gas downstream of the catalyst, and the main feedback correction value DF (output VAF) is corrected to the increase side or the decrease side. A decrease in the accuracy of the main feedback control due to variations in output characteristics or changes with time is suppressed.

[Calculation procedure of sub feedback learning value SG]
The sub feedback learning value SG is calculated using the following equation (13) based on the first learning value A, the second learning value B, and the interpolation correction value K.

SG = B + (A−B) · K (13)
SG: Sub feedback learning value
A: First learning value
B: Second learning value
K: Interpolation correction value In the equation (13), the first learning value A is a sub-feedback as a learning value corresponding to when the maximum lift amount and the operating angle of the intake valve 9 are small (in this example, when it becomes the minimum value). It is updated based on the correction value VH. Specifically, when the maximum lift amount and operating angle of the intake valve 9 are small, the latest sub-feedback correction value VH is subjected to a gradual change process to calculate the update amount SGK. Based on the update amount SGK obtained by performing the lower limit guard, the update is performed using an expression “A ← previous A + SGK (14)”. That is, by adding the updated amount SGK after guard to the previous first learned value A, the first learned value A is updated.

  In Expression (13), the second learning value B is updated based on the sub feedback correction value VH as a learning value corresponding to when the maximum lift amount and the operating angle of the intake valve 9 are large (in this example, when the maximum value is reached). It is what is done. Specifically, when the maximum lift amount and operating angle of the intake valve 9 are large, the latest sub-feedback correction value VH is subjected to a gradual change process to calculate the update amount SGK. Based on the update amount SGK obtained by guarding the lower limit, the update is performed using the formula “B ← previous B + SGK (15)”. That is, by adding the updated amount SGK after guard to the previous second learned value B, the second learned value B is updated.

  In equation (13), the interpolation correction value K is calculated from “0” to “1.0” as shown in FIG. 6 based on the current maximum lift amount and operating angle of the intake valve 9 and the current intake pressure of the engine 1. ”, And determines the sensitivity of changes in the sub-feedback learning value SG with respect to the update of the first learning value A and the second learning value B. Note that the maximum lift amount and operating angle of the intake valve 9 are obtained based on the rotation angle of the electric motor 15 detected by the position sensor 35. As the interpolation correction value K is decreased, the sensitivity of the change in the sub feedback learning value SG with respect to the update of the first learning value A and the second learning value B is decreased.

  As can be seen from the equations (13) to (15), when the sub feedback correction value VH is larger than “0”, the sub feedback learning value SG is updated to the increase side, and the main feedback correction by the learning value SG is performed. The fuel injection amount is increased through correction of the value DF (output VAF) to the increase side. Further, when the sub feedback correction value VH is smaller than “0”, the sub feedback learning value SG is updated to the decreasing side, and the main feedback correction value DF (output VAF) by the learning value SG is decreased. The fuel injection amount is reduced through the correction.

  Through the calculation of the sub feedback learning value SG and the correction of the main feedback correction value DF using the learning value SG as described above, the sub feedback correction value VH comes closer to “0”. Further, the sub-feedback learning value SG when the sub-feedback correction value VH approaches “0” to some extent and is stabilized becomes a value corresponding to a steady deviation of the air-fuel ratio of the engine 1 from the stoichiometric air-fuel ratio. Then, the sub-feedback learning value SG at this time becomes a value learned as a value corresponding to the steady deviation.

  Incidentally, the cause of the steady deviation of the actual air-fuel ratio of the engine 1 from the stoichiometric air-fuel ratio is the connection of the intake passage 3 to the combustion chamber 2 due to assembly errors around the intake valve 9 and deposits deposited on the valve 9. It can also be mentioned that the passage area of the portion deviates from an appropriate value (solid line) by a deviation amount ΔS as indicated by a two-dot chain line in FIG. The steady deviation of the actual air-fuel ratio due to these causes with respect to the theoretical air-fuel ratio is expressed by the distance in the vertical axis direction between the broken line (theoretical air-fuel ratio) and the one-dot chain line (actual air-fuel ratio) in FIG. As the maximum lift amount and operating angle of the intake valve 9 are reduced and the intake air amount is reduced, the intake valve 9 is increased. Accordingly, the steady deviation of the actual air fuel ratio from the theoretical air fuel ratio due to the above causes mainly occurs when the maximum lift amount and the operating angle of the intake valve 9 are small.

  This is because as the maximum lift amount and operating angle of the intake valve 9 are reduced and the passage area is reduced, the ratio of the deviation amount ΔS in FIG. 7 to the entire passage area is increased. In other words, the smaller the maximum lift amount and operating angle of the intake valve 9 and the smaller the intake air amount, the more the ratio of the deviation (corresponding to ΔS) of the intake air amount due to the above causes to the total intake air amount. This is because it becomes larger. As described above, when the maximum lift amount and the operating angle of the intake valve 9 are small, the ratio of the deviation amount to the appropriate value of the intake air amount of the engine 1 due to the above causes increases in the entire intake air amount, which is the actual air-fuel ratio. It appears as a steady deviation from the theoretical air-fuel ratio. On the other hand, when the maximum lift amount and the operating angle of the intake valve 9 are large, the intake air amount of the engine 1 increases, so that the entire intake air amount of the deviation from the appropriate value of the intake air amount of the engine 1 due to the above cause is increased. The occupation ratio becomes small, and it becomes difficult to show it as a steady deviation of the actual air-fuel ratio with respect to the theoretical air-fuel ratio.

  Also for the steady deviation of the actual air-fuel ratio of the engine 1 from the theoretical air-fuel ratio due to the above cause, the sub-feedback learning value SG is learned as a value corresponding to the steady deviation, and the fuel injection amount is based on the learned value SG. It is possible to achieve compensation by correcting. Specifically, when the maximum lift amount and the operating angle of the intake valve 9 are small, the first learning value A of the equation (13) is updated based on the sub feedback correction value VH that changes corresponding to the steady deviation. The Then, by calculating the sub feedback learning value SG using the above equation (13) based on the updated first learning value A and the like, the steady state of the actual air fuel ratio due to the cause of the sub feedback learning value SG with respect to the theoretical air fuel ratio is steady. Learning to the value corresponding to the gap is performed. By compensating the fuel injection amount based on the learned sub-feedback learning value SG, compensation for the steady deviation can be achieved.

  The sub feedback learning value SG is also calculated using an interpolation correction value K that is variably set based on the current maximum lift amount and operating angle of the intake valve 9 and the intake pressure of the engine 1. Therefore, the calculated sub feedback learning value SG can be set to a value corresponding to the maximum lift amount and operating angle of the intake valve 9 and the intake pressure of the engine 1.

  The interpolation correction value K is set to a smaller value as the maximum lift amount and operating angle of the intake valve 9 are increased as shown in FIG. 6, and sub-feedback learning for updating the first learning value A and the second learning value B is performed. The sensitivity of the change in value SG is reduced. This is because the larger the maximum lift amount and the operating angle of the intake valve 9 and the larger the intake air amount, the smaller the ratio of the deviation amount with respect to the appropriate value of the intake air amount due to the above cause to the entire intake air amount. This is because the influence on the steady deviation of the actual air-fuel ratio from the actual air-fuel ratio due to is reduced, and the sub-feedback learning value SG is learned accordingly.

  Further, with respect to the interpolation correction value K, when the throttle valve 13 is driven to the closed side in order to reduce the intake air amount at the time of engine low load operation or when the variable valve lift mechanism 14 is stuck, the intake of the engine 1 is reduced. The value decreases as the atmospheric pressure decreases, and the sensitivity of the change in the sub-feedback learning value SG to the update of the first learning value A and the second learning value B is reduced. This is because, as the throttle valve 13 is driven to the closed side and the intake pressure of the engine 1 decreases, the actual air-fuel ratio steadily deviates from the theoretical air-fuel ratio in the deviation from the appropriate value of the intake air amount of the engine due to the above-mentioned cause. This is because the sub-feedback learning value SG is learned so as to reduce the influence on the sub-feedback.

  By calculating the interpolation correction value K as described above and calculating the sub feedback learning value SG using the above (13) based on the interpolation correction value K, the following effects can be obtained. That is, the engine is operated in a state where the maximum lift amount and operating angle of the intake valve 9 and the opening of the throttle valve 13 are adjusted to values different from the values when the first learning value A and the second learning value B are updated. In such a situation, the sub-feedback learned value SG becomes an appropriate value as a value corresponding to a steady deviation of the actual air-fuel ratio from the theoretical cause with respect to the theoretical air-fuel ratio.

  FIG. 9 shows transitions with respect to changes in the maximum lift amount and operating angle of the intake valve 9 and changes with respect to changes in the intake pressure of the engine 1 at the sub-feedback learning value SG. As can be seen from the figure, the first learning value A and the second learning value B used for calculating the sub-feedback learning value SG are the maximum and the maximum when the maximum lift amount and the operating angle of the intake valve 9 are the minimum values, respectively. The value corresponds to the value. The value obtained by subtracting the second learning value B from the first learning value A (“A−B”) is variable according to the current maximum lift amount and operating angle of the intake valve 9 and the current intake pressure of the engine 1. The sub-feedback learning value SG is calculated by multiplying the set interpolation correction value K and adding the multiplied value “(A−B) · K” to the second learning value B. As with the interpolation correction value K, the sub-feedback learning value SG calculated in this way becomes a smaller value as the maximum lift amount and operating angle of the intake valve 9 become larger, and the throttle valve 13 is driven to the closing side and the engine 1 The value becomes smaller as the intake pressure decreases.

Next, a problem that occurs in relation to the calculation of the sub feedback learning value SG will be described.
In an operating state where the required amount of intake air in the engine 1 is small, such as during low engine load operation, the intake air amount can be reduced to the required amount only by changing the maximum lift amount and operating angle of the intake valve 9 by driving the variable valve lift mechanism 14. Since the amount cannot be reduced, the maximum lift amount and operating angle of the intake valve 9 are reduced, and the throttle valve 13 is further driven to the closing side. In addition, even when the maximum lift amount and the operating angle of the intake valve 9 are decreased due to the sticking abnormality of the variable valve lift mechanism 14 and the intake air amount cannot be reduced to the required amount, the intake air amount is reduced to the required amount. The throttle valve 13 is driven to the closing side in order to realize the weight reduction.

  When the maximum lift amount and the operating angle of the intake valve 9 are reduced and the throttle valve 13 is driven from the fully open state to the closed side, the sub feedback feedback learning value SG is set to the actual sky due to the above cause by updating the first learning value A. If the learning is performed so as to have a value corresponding to a steady deviation from the stoichiometric air-fuel ratio, the following problems may occur. That is, the learned sub-feedback learning value SG may become a low-accuracy value as a value corresponding to the steady deviation of the actual air-fuel ratio from the theoretical air-fuel ratio due to the above cause. In the variable range of the maximum lift amount and the operating angle of the intake valve 9 shown in FIG. 9 and the change region of the intake pressure of the engine 1, the region where the above-described problem occurs is shown in FIG. This is a region R1 surrounded by a dotted line.

  Here, there is a steady deviation of the actual air-fuel ratio with respect to the theoretical air-fuel ratio due to the above cause, and when the maximum lift amount and the operating angle of the intake valve 9 are small, the sub feedback learning value SG corresponds to the steady deviation. Therefore, it is important to perform the learning in a state where the throttle valve 13 is almost fully opened in order to learn with high accuracy so as to obtain the value to be achieved. Therefore, the learning of the sub feedback learning value SG by updating the first learning value A is performed in the state where the throttle valve 13 is driven to the closed side as described above and the maximum lift amount and the operating angle of the intake valve 9 are small. When it is performed, it becomes difficult to learn the sub-feedback learning value SG to be a value corresponding to the steady deviation.

  This is because when the throttle valve 13 is driven to the closed side rather than the fully opened state when the maximum lift amount and the operating angle of the intake valve 9 are small, the throttle valve opening for adjusting the intake air amount of the engine 1 is performed. This is related to an increase in the share due to the adjustment of the degree. That is, when the intake air amount of the engine 1 is adjusted, the share of the throttle valve 13 for adjusting the opening is increased, and the maximum lift of the intake valve 9 for adjusting the intake air amount is increased. The share by adjusting the amount and the operating angle is reduced. Thus, when the share by adjustment of the maximum lift amount and the operating angle of the intake valve 9 for adjusting the intake air amount of the engine 1 is reduced, the actual deviation of the actual air-fuel ratio from the theoretical air-fuel ratio due to the above cause is reduced. The impact on is reduced. Therefore, the first learning value A is updated based on the sub-feedback correction value VH so that the sub-feedback learning value SG becomes a value corresponding to the steady deviation, thereby learning the sub-feedback learning value SG. However, it becomes difficult to make the learned value SG a highly accurate value.

  In order to suppress such a decrease in accuracy of the sub-feedback learning value SG, the throttle valve 13 is driven from the fully opened state to the closed side, and the determination value is determined so that the intake pressure of the engine 1 is lower than the intake pressure when the throttle valve is fully opened. It is conceivable that learning of the sub-feedback learning value SG is prohibited when it decreases below. In FIG. 9, the intake pressure of the engine 1 when the throttle valve 13 is fully opened is shown in each solid line showing the transition of the sub-feedback learning value SG accompanying the change in the maximum lift amount and the operating angle of the intake valve 9. The corresponding transition is indicated by a solid line located on the uppermost side, and the corresponding transition when the intake pressure reaches the determination value is indicated by a solid line H in the figure.

  Accordingly, learning of the sub-feedback learning value SG is prohibited in the region below the solid line H in FIG. 9, and learning of the sub-feedback learning value SG in the region R1 in FIG. Learning of the sub-feedback learning value SG by updating based on the sub-feedback correction value VH of A is also prohibited. By prohibiting learning of the sub-feedback learning value SG in this way, the learning of the sub-feedback learning value SG by updating the first learning value A when the throttle valve 13 is driven from the fully open state to the closing side results in the same. A decrease in accuracy as a value corresponding to a steady deviation of the actual air fuel ratio from the theoretical air fuel ratio due to the above cause in the learning value SG is suppressed.

  However, when the intake pressure of the engine 1 is equal to or lower than the determination value, that is, when learning of the sub feedback learning value SG is prohibited in the region below the solid line H in FIG. 9, sub feedback by updating the first learning value A is performed. Learning of the learning value SG and learning of the sub-feedback learning value SG by updating the second learning value B are both prohibited. In other words, not only the learning of the sub feedback learning value SG by the update of the first learning value A is prohibited, but also the sub of the second learning value B corresponding when the maximum lift amount and the operating angle of the intake valve 9 are large. Learning of the sub feedback learning value SG by updating the feedback correction value VH is also prohibited. For this reason, it is unavoidable that the learning frequency of the sub-feedback learning value SG decreases. When the learning frequency of the sub-feedback learning value SG decreases in this way, when the steady deviation of the actual air-fuel ratio from the theoretical air-fuel ratio due to a cause other than the above causes occurs, the second learning value B is updated. Since learning of the sub-feedback learning value SG is prohibited, it takes time until the learning value SG is learned so as to be a value corresponding to the deviation.

  Further, when the maximum lift amount and the operating angle of the intake valve 9 are fixed at the values at the time of medium load operation, the adjustment of the intake air amount of the engine 1 to the required amount at the time of low load operation is performed on the closing side of the throttle valve 13. Realized by driving to. For this reason, the intake pressure of the engine 1 tends to be equal to or less than the determination value. In other words, the solid line indicating the transition of the sub-feedback learning value SG accompanying the change in the maximum lift amount and the operating angle of the intake valve 9 tends to be lower than the solid line H in FIG. For this reason, learning of the sub-feedback learning value SG is likely to be prohibited, and the learning frequency of the learning value SG is more likely to be lowered. As a result, the sub-feedback learning value SG is used for correcting the fuel injection amount without being learned so that the sub-feedback learning value SG becomes a value corresponding to the above-described steady deviation, whereby the actual air-fuel ratio of the engine 1 is reduced. There is a risk that effective exhaust purification by the three-way catalyst cannot be performed due to the deviation from the theoretical air-fuel ratio.

Next, the countermeasure of this embodiment regarding the malfunction which arises in connection with calculation of the said sub feedback learning value SG is demonstrated.
FIG. 10 shows an update routine for determining whether learning is performed by updating the first learning value A or by updating the second learning value B when learning the sub-feedback learning value SG. It is a flowchart. This update routine is periodically executed by, for example, a time interruption every predetermined time when the sub feedback learning value SG is learned.

  In this routine, first, an interpolation correction value K is calculated as shown in FIG. 6 based on the maximum lift amount and operating angle of the intake valve 9 and the intake pressure of the engine (S101), and then the interpolation correction value K is calculated. Is less than a predetermined determination value J (S102). This determination value J is set to a value at or near the center of the frequently used range of the variable setting range from “0” to “1.0” of the interpolation correction value K.

  When the interpolation correction value K is greater than or equal to the determination value J, an update based on the sub feedback correction value VH of the first learning value A corresponding to the maximum lift amount and operating angle of the intake valve 9 is small (S103). The sub feedback learning value SG is learned based on the update of the first learning value A. When the interpolation correction value K is less than the determination value J, the second learning value B corresponding to when the maximum lift amount and the operating angle of the intake valve 9 are large is updated based on the sub feedback correction value VH (S104). ), The sub feedback learning value SG is learned based on the update of the second learning value B.

  Thus, by performing learning of the sub-feedback learning value SG in a different manner depending on the magnitude of the interpolation correction value K, the occurrence of the above-described problems can be avoided. Hereinafter, this reason will be described with reference to FIG.

  FIG. 11 shows the transition with respect to changes in the maximum lift amount and operating angle of the intake valve 9 and the intake pressure of the engine 1 at the interpolation correction value K, as in FIG. 6, and the interpolation correction value K and the determination value J Shows the relationship. Further, a region R2 surrounded by a two-dot chain line in FIG. 11 is a region corresponding to the region R1 shown in FIG. Therefore, also in the region R2, as in the region R1, the sub feedback learning value SG corresponds to the steady deviation based on the update of the first learning value A when the maximum lift amount and the operating angle of the intake valve 9 are small. This is a region that may be a low precision value.

  As described above, only when the interpolation correction value K is greater than or equal to the determination value J, the first learning value A is updated based on the sub-feedback correction value VH, thereby subtracting the sub-feedback learning value SG from the actual air-fuel ratio due to the above cause. Learning of the learning value SG is performed to obtain a value corresponding to a steady deviation with respect to the theoretical air-fuel ratio. The reason why the interpolation correction value K becomes larger than the judgment value J is that the throttle valve 13 is fully opened or almost fully opened, the intake pressure of the engine 1 is increased, and the maximum lift amount and operating angle of the intake valve 9 are decreased. Thus, the interpolation correction value K becomes large.

  Accordingly, when the maximum lift amount and the operating angle of the intake valve 9 are small and the throttle valve 13 is driven to the closed side, that is, in the region R2 in the figure, the same learning by updating the first learning value A is performed. Learning of the value SG is never performed. Then, by learning such a sub-feedback learning value SG, the accuracy as a value corresponding to a steady deviation of the actual air-fuel ratio of the engine 1 with respect to the theoretical air-fuel ratio due to the above cause in the sub-feedback learning value SG. It is suppressed that falls.

  On the other hand, when the interpolation correction value K is less than the determination value J, as described above, the second learning value B is updated based on the sub-feedback correction value VH, whereby the sub-feedback learning value SG is changed to the theoretical air-fuel ratio of the actual air-fuel ratio. Learning of the learning value SG is performed to obtain a value corresponding to a steady deviation with respect to. It should be noted that the interpolation correction value K is less than the determination value J when the maximum lift amount and the operating angle of the intake valve 9 are increased, thereby reducing the interpolation correction value K. Even if the maximum lift amount and the operating angle of the intake valve 9 are small, the interpolation correction value K may be less than the determination value J when the throttle valve 13 is driven to the closed side and the intake pressure of the engine 1 becomes low. Is expensive.

  Accordingly, when the maximum lift amount and operating angle of the intake valve 9 are small and the throttle valve 13 is driven to the closed side, that is, in the region R2 in the figure, the maximum lift amount and operating angle of the intake valve 9 are set. It is possible to learn the sub-feedback learning value SG by updating the second learning value B corresponding to a large value. Therefore, in the state as described above, that is, in the region R2, both the learning of the sub feedback learning value SG by updating the first learning value A and the learning of the sub feedback learning value SG by updating the second learning value B are both performed. This prohibits the reduction of the learning frequency of the learning value SG. Specifically, when there is a steady deviation of the actual air-fuel ratio of the engine 1 from the theoretical air-fuel ratio due to a cause other than the above-mentioned causes, the second learning value B cannot be updated by prohibiting the learning, so that A decrease in the learning frequency of the sub-feedback learning value SG is suppressed. As a result, there is no problem that it takes time until the sub feedback learning value SG is learned so as to have a value corresponding to the steady deviation.

According to the embodiment described in detail above, the following effects can be obtained.
(1) When the interpolation correction value K is greater than or equal to the determination value J, learning of the sub-feedback learning value SG based on the update of the first learning value A is performed, and when the interpolation correction value K is less than the determination value J, second learning is performed. The sub feedback learning value SG based on the update of the value B is learned. Thereby, it is possible to suppress the learning accuracy from being lowered due to the learning of the sub feedback learning value SG based on the update of the first learning value A in the state where the throttle valve 13 is driven to the closed side, and the second in the above state. The learning of the sub-feedback learning value SG based on the update of the learning value B can suppress a decrease in the learning frequency of the learning value SG. Accordingly, the sub-feedback learning value SG is reflected in the main feedback correction value DF, and the fuel injection amount is corrected by the correction value DF to compensate for a steady deviation between the actual air-fuel ratio of the engine 1 and the theoretical air-fuel ratio. Thus, the compensation can be performed accurately.

  (2) The determination value J is set to a value at or near the center of the frequently used range of the variable setting range from “0” to “1.0” in the interpolation correction value K. For this reason, there is a high possibility that the opportunity for the interpolation correction value K to be equal to or greater than the determination value J and the opportunity for the interpolation correction value K to be less than the determination value J to occur equally. Accordingly, the frequency at which the sub feedback learning value SG is learned by updating the first learning value A when the interpolation correction value K is equal to or greater than the determination value J, and the second learning is performed when the interpolation correction value K is less than the determination value J. It is possible to equalize the frequency with which the sub feedback learning value SG is learned by updating the value B.

In addition, the said embodiment can also be changed as follows, for example.
The determination value J can be set to a value other than the value in the center or near the center of the frequently used range of the variable setting range from “0” to “1.0” in the interpolation correction value K.

  As for the interpolation correction value K, a value multiplied by the difference (“A−B”) between the first learning value A and the second learning value B is exemplified, but instead, it is added to the difference. It is also possible to adopt a value.

  As the first learning value A, instead of adopting the corresponding values when the maximum lift amount and the operating angle of the intake valve 9 become the minimum values, to some extent than when the maximum lift amount and the operating angle become the minimum values A value corresponding to a large value may be adopted. In this case, the interpolation correction value K is also changed corresponding to such a change.

  As the second learning value B, instead of adopting the corresponding values when the maximum lift amount and the operating angle of the intake valve 9 become the maximum values, a certain amount is obtained than when the maximum lift amount and the operating angle become the maximum values. A value corresponding to a small value may be adopted. In this case, the interpolation correction value K is also changed corresponding to such a change.

  DESCRIPTION OF SYMBOLS 1 ... Engine, 2 ... Combustion chamber, 3 ... Intake passage, 4 ... Fuel injection valve, 5 ... Spark plug, 6 ... Piston, 7 ... Crankshaft, 8 ... Exhaust passage, 9 ... Intake valve, 10 ... Exhaust valve, 11 DESCRIPTION OF SYMBOLS ... Intake camshaft, 12 ... Exhaust camshaft, 13 ... Throttle valve, 14 ... Variable valve lift mechanism, 15 ... Electric motor, 16 ... Catalytic converter, 17 ... Air-fuel ratio sensor, 18 ... Air-fuel ratio sensor, 21 ... Electronic control device , 27 ... accelerator pedal, 28 ... accelerator position sensor, 30 ... throttle position sensor, 32 ... air flow meter, 33 ... intake pressure sensor, 34 ... crank position sensor, 35 ... position sensor.

Claims (3)

A variable valve lift mechanism that adjusts the intake air amount by varying the maximum lift amount and operating angle of the intake valve that opens and closes to open / close the communication between the intake passage and the combustion chamber, and is provided in the intake passage and is fully open during normal operation. On the other hand, when the amount of intake air cannot be reduced only by driving the variable valve lift mechanism, a throttle valve that is driven to the closed side rather than the full open, and a signal that is provided in the exhaust passage and corresponds to the oxygen concentration in the exhaust gas A feedback correction value that is applied to an internal combustion engine that includes a sensor that outputs a sensor, and that increases or decreases based on a deviation from a target value, which is a value when the actual air-fuel ratio of the engine at the same output is the stoichiometric air-fuel ratio. And a feedback learning value learned to be a value corresponding to a steady deviation of the actual air-fuel ratio from the theoretical air-fuel ratio based on the feedback correction value that increases or decreases And by reflecting the fuel injection amount of the internal combustion engine, air-fuel ratio control apparatus for an internal combustion engine for controlling the actual air-fuel ratio of the internal combustion engine to the stoichiometric air-fuel ratio,
The feedback learning value includes a first learning value corresponding to when the maximum lift amount and operating angle of the intake valve are small, a second learning value corresponding to when the maximum lift amount and operating angle of the intake valve is large, It is calculated using an interpolation correction value that is variably set based on the current maximum lift amount and operating angle of the intake valve and the intake pressure of the current internal combustion engine,
The interpolation correction value is a maximum lift of the intake valve so that the feedback learning value calculated based on the interpolation correction value becomes a value corresponding to the current maximum lift amount and operating angle of the intake valve and the current intake pressure. As the amount and operating angle increase and the intake pressure decreases, the feedback learning value is decreased.
The learning of the feedback learning value for setting the feedback learning value to a value corresponding to a steady deviation of the actual air-fuel ratio from the stoichiometric air-fuel ratio is performed when the interpolation correction value is equal to or greater than a predetermined determination value. It is realized by updating the first learning value based on the feedback correction value, and is realized by updating the second learning value based on the feedback correction value when the interpolation correction value is less than the determination value. An air-fuel ratio control apparatus for an internal combustion engine characterized by The first learning value is a value corresponding to when the maximum lift amount and the operating angle of the intake valve become a minimum value,
The second learning value is a value corresponding to the maximum lift amount and operating angle of the intake valve reaching a maximum value,
The interpolation correction value is variable in the range of “0” to “1.0” based on the maximum lift amount and operating angle of the intake valve and the intake pressure, and the larger the maximum lift amount and operating angle, the “ As the intake pressure decreases, the value decreases toward “0”.
The feedback learning value is the following when the learning value is “SG”, the first learning value is “A”, the second learning value is “B”, and the interpolation correction value is “K”. Is calculated using the equation “SG = B + (A−B) · K”,
2. The internal combustion engine according to claim 1, wherein the determination value is set to a value at or near a center of a frequently used range of a variable setting range from “0” to “1.0” in the interpolation correction value. Air-fuel ratio control device. The internal combustion engine has a plurality of cylinders,
The sensor is a sensor provided between a collection portion of exhaust passages connected to the cylinders and a catalyst downstream of the exhaust passage, and a sensor provided downstream of the catalyst,
The feedback correction value increases or decreases based on the deviation between the output of the sensor downstream of the catalyst and the target value, and the actual air-fuel ratio of the internal combustion engine is calculated based on the output of the sensor upstream of the catalyst. Is a sub-feedback correction value that is calculated as a correction value for the fuel injection amount and reflected in the main feedback correction value that is used to correct the fuel injection amount,
The feedback learning value is learned based on the sub-feedback correction value to be a value corresponding to a steady deviation of the actual air-fuel ratio of the internal combustion engine from the stoichiometric air-fuel ratio, and is reflected in the main feedback correction value. The air-fuel ratio control apparatus for an internal combustion engine according to claim 1 or 2, wherein the learning value is a learning value.

Images